CN104380306A - Real time posture and movement prediction in execution of operational tasks - Google Patents

Abstract

A system for determining the pose of an articulated model of a virtual subject in carrying out a task within a vehicle occupant packaging design is described. The system uses an initial posture of the articulated model prior to the virtual subject carrying out the task as a starting point. The system determines pose throughout the carrying out of the task by the virtual subject. The pose is determined based on the parameters of the virtual subject, the design, the task to be completed, and a set of constraints restricting the motion of the virtual subject. The pose can be analyzed to determine the feasibility of a design for a human subject, without the need for live subject testing. The method is analytically derived and results in a kinematically and dynamically consistent posture in real-time without requiring iterative optimization.

Description

Real-time attitude when executable operations task and moving projection

Technical field

Present disclosure mainly relate to in vehicle for determine the radial type model of the availability of vehicle occupant package design posture prediction field.

Background technology

Vehicle occupant packaging refers to the part taken by the driver of vehicle and passenger of the inner space of vehicle.Vehicle occupant packaging can comprise multiple different characteristic, such as, comprise, and seat designs, rim brake location and operation, the design of bearing circle locating and orienting, central control board and door handle design and operate.Vehicle occupant is packed to relate to and is referred to following general field, and this general field relates to design vehicle occupant packaging, thus the inside of given vehicle not only practicality but also comfortable.Change widely due to Car design and improve with being iterated along with every vehicle of new generation, also needing on continuing basis, redesign and improve vehicle occupant packaging.

Usually, by generating the full scale model of given design, and then utilize multiple different people human subject to test this design to test new vehicle occupant packaging.By the human subject crossing over the extensive physical characteristics such as comprising height, body weight, sex, limbs length (such as leg and arm lengths), intensity and range of motion ideally and launch to use in testing.This helps the vehicle occupant packaging ensureing test and approval can be operated by the overwhelming majority of human colony.

Some different software simulation package are available, these software emulation bags permission emulation to the test of virtual subject to the emulation of vehicle occupant package design and permission.These virtual subjects are computer models of human subject, and wherein virtual subject has the identical change of the physical characteristics (such as height, body weight, limbs length) used in the vehicle package design test of actual life.The example of these software packages such as comprises the DELMIA of JACK and the DASSAULT SYSTEMS supply of SIEMENS supply.

These software packages improve vehicle occupant package design process and without the need to the physical prototype for each design iteration by allowing design iteration.Such as, whether Software for Design bag allows deviser's test person human subject by match in given design (such as, whether they will can reach rim brake physically within the scope of its whole full motion).In addition, these software packages allow to calculate operate vehicle certain in (such as, the attitude of fixing in time when grasping rim brake) time single static mankind's attitude.Generally speaking, these software package Using statistics return and calculate single static attitude, and these statistical regressions need to calculate for a long time, therefore can not be determined in real time.

The shortcoming of existing design software bag is that they can not be provided in the FR information of collecting when carrying out the test of living person's human subject with full scale test model at present.Thus, test is lived so that the shortcoming making up existing Software for Design bag remains common practice to the design of finalize a text (or half finalizes a text).

Summary of the invention

Embodiments of the invention provide a kind of method (and correspondence system and computer program) of the posture for determining the virtual subject in the vehicle occupant package design being limited by constrain set.

In one embodiment, describe a kind of computer-implemented method for determining posture, system and comprise the non-transient computer-readable recording medium of executable computer program code.Determine that posture comprises the initial attitude of access radial type system at initial time, initial attitude comprises the angle of the multiple degree of freedom for radial type system.Determine the task that attitude also comprises access vehicle occupant package design and will be realized by radial type system in this design.Determine that posture also comprises the constrain set of accessing and limiting to realize task the degree of freedom how handling radial type system, wherein constraint comprises the contiguity constraint of the motion away from the parts be present in design preventing radial type system from carrying out.During realizing task, after initial time, determine multiple attitudes of radial type system, wherein determine that attitude is included in also in accordance with the degree of freedom manipulation in time determining the radial type system of the task that realizes while constrain set.

The feature and advantage described in the description are not exhaustive, and specifically, many supplementary features and advantage will in view of accompanying drawing, instructions and claim by those of ordinary skill in the art be clear.In addition, should be noted that the speech that uses in the description is main for readable and instruct object and selected and can not be selected for define or limit disclosed subject content.

Accompanying drawing explanation

Fig. 1 is the block diagram of diagram according to the computer system for assessment of vehicle occupant package design of an embodiment.

Fig. 2 be according to during the executing the task in vehicle tenant package design an of embodiment for determining the process flow diagram of virtual subject attitude in time.

Fig. 3 is the example diagram of the radial type model of virtual subject according to an embodiment.

Fig. 4 be according to an embodiment for the process flow diagram of indivedual attitudes during determining to realize task in design.

Each figure only describes the various embodiments of embodiment for exemplary purposes.Those skilled in the art easily will recognize the alternative can using structure shown in this article and method and the principle not departing from embodiment described herein from following discussion.

Embodiment

Referring now to each figure, embodiment is described, the unit that wherein similar label instruction is identical or functionally similar.Also in the various figures, the leftmost bit of each label corresponds to the figure that wherein first time uses this label.

System survey

Fig. 1 is the block diagram of diagram according to the computer system 100 for assessment of vehicle occupant package design of an embodiment.Generally speaking, computer system 100 receives one or more physical tasks (also referred to as operation task) that vehicle occupant package design (referred to as design) to be assessed, the parameter describing the radial type model of virtual subject, the constrain set limiting the motion of virtual subject in design and virtual subject will perform in design.Posture during virtual subject performs in vehicle occupant package design at virtual subject in physical tasks one or more physical tasks that computer system 100 is configured to determine (or tracking).The posture of analyzing virtual main body when executing the task is to determine the feasibility (or availability) of the design for the potential driver of mating with the size and dimension of virtual subject and/or passenger.

Design describes the inside passenger cabin of vehicle.Design comprises multiple parts, and the example of these parts comprises seat, headrest, bearing circle, pedal (such as gasoline, brake and clutch coupling), rim brake, the audio-frequency/video frequency system being arranged in central control board, instrument bar (such as controlling lamp and wiper) and the instrument panel with length and height.This component enumerates is only example and has no intention exhaustive.This design also comprises size (such as ratio) for parts and the relative distance between various parts, absolute fix and orientation.Such as, the distance between bearing circle and seat and between pedal and seat can be also included in the design.Design the scope that may locate that also can comprise for various parts.Such as, in many designs, can raise in the framework of passenger cabin as a whole or reduce, tilt or forward or backward sliding seats.Similarly, can move forward or backward or raise or reduce bearing circle.Can reorientate these parts and redirect affects particular vehicle occupant package design in large quantities by the availability of the different segment of human colony.

Virtual subject is designated the radial type model of actual persons human subject by computer system 100.By carrying out modeling according to radial type model to human subject, computer system 100 allows the assessment designed without the need to full scale model and mankind's test subject.Generally speaking, the radial type model of virtual subject is similar, because the majority of human colony has two arms, two legs, trunk, head, neck, waists etc.The parameter of virtual subject allows the differentiation between virtual subject, and this differentiation is reflected in the differentiation between the member of human colony as a whole.Limbs length that parameter can comprise (such as forearm, upper arm, thigh and shank and torso length), altogether virtual subject height, virtual subject general assembly (TW), range of motion, the virtual subject vision visual field, deformity and further feature.As described above, this parameter list is only example and has no intention exhaustive.

Fig. 3 is the example diagram of the radial type model of virtual subject according to an embodiment.In the example of fig. 3, virtual subject is by the multiple characterizing definitions on health, and these features comprise the crown, right and left shoulder, right and left elbow, waist, right and left wrist, right and left hip, right and left knee and right and left ankle.Generally speaking, feature be positioned at the joint that can rotate around one or more axle or near.The axle that joint can rotate around it is called degree of freedom.Given joint can have more than one degree of freedom.Such as, mankind's elbow can rotate, therefore have two degree of freedom around two axles.One degree of freedom associates with flexing/extension, and the second degree of freedom associates with outer turning with interior turn.Generally speaking, the angle of the degree of freedom of virtual subject and parameter fully specify the static immobilization of all limbs of virtual subject.This is also referred to as attitude.

In one implementation, the representative of the parameter for virtual subject one or more thresholding in human colony as a whole of reception.Such as, the parameter for virtual subject of reception can represent for height and/or body weight and/or limbs length and/or about certain other standard be percent 50,75,90 or 95 driver or passenger.The virtual subject that assessment represents one of these thresholdings allows computer system 100 to determine the feasibility of Car design with regard to the ratio of colony.Such as, can represent according to being highly human colony's percent 5 and 95 for the parameter of two different virtual main bodys.Computer system 100 can about these two virtual subject assessment designs.If design is feasible for these two virtual subjects, then computer system 100 can infer design for human colony to drop on according to the whole part in percent 5 and 95 of height be feasible.The test design relatively meeting the virtual subject of various thresholding improves the efficiency of design test by the nonessential virtual subject avoiding testing in the scope that drops on after tested.The virtual subject that test represents thresholding also allows computer system 100 to report to estimate that the similar feasibility of test result finds with industrial.

Task is set forth the target that realizes by the motion of virtual subject in design.Task such as can comprise the manipulation (such as, pulling rim brake) of one or more parts to design.In one implementation, task can set forth the concrete motion path will followed to realize task.When virtual subject reaches the end in the path of specifying, task has been considered as.Can use the path of specifying, wherein how design regulation itself can handle some parts in design.Use above-mentioned rim brake example, when design can specify in and pull rim brake, rim brake only can be advanced through certain path, such as produces angle relative to point of fixity and rotates.In other example, except designated movement path, task also can only be specified for the starting point of task and terminal the motion of the sun sunshading board on the top, top of the side window covering driver (be such as used for).In these examples, followed the tracks of the posture of virtual subject by the possible path of reach home in many possible paths.When reaching home, task has been considered as.

How constrain set restriction virtual subject can be mobile in design when realizing task.Constrain set can comprise some dissimilar constraints, and these constraints such as comprise one or more contiguity constraint, one or more uncomfortable constraint, one or more joint constraints retrains, constraint is avoided in one or more collision and dynamic conformance constraint.Such as, contiguity constraint can be specified in the whole process realizing task, to maintain contact between the thigh of main body and/or lazyback to indicate virtual subject.Another contiguity constraint can be defined to maintain the contact etc. between the pin of main body and the pedal of automobile.

In one implementation, computer system 100 comprises attitude initialization system 102, posture certainty annuity 104 and design analysis system 106.

Attitude and heading reference system 102 is configured to use design, the incompatible initial attitude determining virtual subject of virtual subject parameter, task to be done and constraint set.As introduced above, attitude refers to the static posture of virtual subject in the special time moment.In one implementation, attitude comprises value vector, and each value describes the orientation (or angle) of radial type model in the degree of freedom in this moment time of virtual subject.From proper the initial attitude of time point determination main body before any task in task to be done.Such as, if task to be done is the rim brake pulling vehicle, then determine the initial attitude of virtual subject, but thus virtual subject allows their hand not yet start drag on rim brake.

In one embodiment, task to be done specifies the starting condition being used for the attitude of virtual subject before beginning task.Use above rim brake example, these starting condition can comprise specifies the hand of virtual subject just should be located in where before beginning task.If virtual subject even can not meet the starting condition (saying nothing of task) of task, then computer system 100 can exit this process and indicate and can not complete the task of specifying for the virtual subject of specifying.Below further describe initial attitude to determine.

Use initial attitude, posture certainty annuity 104 is configured to determine the posture of virtual subject when virtual subject performs the task that one or more specifies when also observing constrain set.Posture refers to virtual subject dynamic (such as changing in time) attitude in the whole process realizing task.In one implementation, posture is included in multiple indivedual attitude that sequential time unit catches.Posture certainty annuity 104 is configured to be received in initial attitude before beginning task, the parameter of virtual subject, task to be done and the constrain set that limits the motion path adopted to realize task as input.If virtual subject can not complete the task of specifying and not violate constraint, then computer system 100 can exit this process and pointer can not complete the task of specifying to the virtual subject of specifying.Below be described in the posture realized in the whole process of one or more task to determine.

Design analysis system 106 is configured to the posture of analysis posture certainty annuity 14 generation to determine the feasibility designed.The feasibility of design can based on multiple Different factor.These factors such as can comprise the amount of energy that virtual subject is made great efforts for the physiology that the uncomfortable numerical degree realizing the quantity of a task (or realizing some tasks) and the torque of needs and/or power, virtual subject is experienced when realizing task, virtual subject maintain to keep static posture when finishing the work simultaneously and the task that realizes from initial attitude consumes.

Based on this analysis, feasibility can be formulated for Yes/No answer, whether this answer instruction virtual subject can complete the task of specifying in the design when also observing constrain set.Also can according to one or more numerical value formulate feasibility, this one or more numerical example as instruction finish the work required torque or power quantity, when finishing the work experience uncomfortable degree, keep task end location time physiology make great efforts and when finishing the work consume energy.These numerical value jointly can represent feasibility and/or closed form can be used to resolve function to be combined into single number to provide single feasibility value.

Initial attitude is determined

As introduced above, attitude initialization system 102 is configured to determine that main body is starting the initial attitude before task to be done when being limited by the constrain set applied the motion of virtual subject.In one embodiment, iterative numerical multiple-objection optimization (MOO) technology is used to determine initial attitude.Initial attitude is based on the parameter of virtual subject.This parameter comprises anthropological measuring parameter, such as limbs yardstick, limbs quality, limbs inertia, limbs center of gravity etc.These parameters specify the ratio of various limbs.

Use this technology, system 102 virtual subject design in when before beginning task at t=t _oexport initial attitude as vector q.Vector q comprises the numerical value for each degree of freedom in the radial type model of virtual subject.That is, initial attitude is described in the orientation in each joint in the health of virtual subject.Vector q is impliedly the function of parameter, and the recurrence of the ratio Using statistics of limbs is derived from parameter.Initial attitude is in the reference frame of design, and therefore initial attitude describes virtual subject and how to be located in vehicle cabin.Can describe quantitatively example initial attitude be virtual subject to be sitting in vehicle seat them arm on the steering wheel with their pin on pedal.

In order to determine initial attitude, system 102 finds vector q, and this vector is the local minimum of scalar function f (q) when being limited by constrain set c (q) about the allowed orientation for each degree of freedom in q.For indivedual degree of freedom q, this can be represented as:

$\underset{q}{\min }f\left(q\right)$

s.t.u _lb≤c(q)≤u _ub.

q _min≤q≤q _max

Function f (q) comprises f1 (q) and these two separate targets of f2 (q), thus f (q)=f1 (q)+f2 (q).First object function f 1 (q) be minimized in the feature (such as their hand and pin) of virtual subject as degree of freedom vector q and parameter specify when prelocalization and virtual subject feature in order to start that task should be in as the distance between the location of specifying in vector p.In one embodiment, minimize based on realizing this to a square summation for tracking error norm:

$f1\left(q\right)=\frac{1}{2}\underset{i=1}{\overset{k}{\mathrm{\Σ}}}{\mathrm{\β}}_i{\left|\right|e_i\left|\right|}^2---\left(2\right)$

In this case, e _iit is the tracking error for each entry in task vector p.Tracking error for individual tasks can mainly describe represent positioning error and the orientation error represented.

$e_i={\left[\begin{array}{cc}{e}_{o_i}& e_{p_i}\end{array}\right]}^T---\left(3\right)$

Definition positioning error is wherein and p _icorrespond respectively to the location of hope for task and prediction.Definition according to the orientation error of angle and axis error is:

$e_o=\frac{1}{2}(n\×n_r+s\×s_r+a\×a_r)---\left(4\right)$

Wherein and R _ithe unit vector ternary of hope and prediction that=[n s a] corresponds respectively to task p represents.The location of the hope of task and the orientation of hope are the parts (or determining by measuring) of design.The location of the prediction of task and the orientation of prediction are the functions of the vector q calculated.Propulsion function is used to determine the locating and orienting of the prediction of task.In addition, β _iit is the scalar giving relative priority for the execution for each task in pending task.

The uncomfortable constraint of the second minimized target equation as described further below 23 definition.This make initial attitude preferentially towards for user more comfortable joint location.

$f2\left(q\right)=\frac{1}{2}{\left|\right|h_1\left(q\right)\left|\right|}^2---\left(2\right)$

Minimizing of function f (q) is limited by by u _lb(lower bound) and u _ubconstraint function c (q) that (upper bound) is defined.Therefore, f (q) is being always maintained at u _lbwith u _ubbetween c (q) when be minimized.U _lband u _ubvalue can be finite or infinite.In one embodiment, constraint function c (q) comprises c1 (q) and these two parts of c2 (q), thus c (q)=c1 (q)+c2 (q).The joint torque τ that first constraint function c1 (q) applies when static equilibrium the joint of virtual subject _scarry out constraint.

τ _s＝c ₁(q)＝τ _g(q)+J ^Tf _es(6)

Wherein f _esto the external force that the joint of virtual subject operates under static condition, and wherein τ _gq () describes the gravitational torque that can operate from the joint to virtual subject of following formula calculating,

${\mathrm{\τ}}_g\left(q\right)=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{m}_j{g}^T{J}_{\mathrm{cogj}}---\left(7\right)$

Wherein expression is at the Jacobi matrix of the center of gravity of every section, and g is 3 × 1 vectors of acceleration of gravity.

Second constraint function c2 (q) is used for avoiding self collision and the collision with environment.In one implementation, c ₂(q)=d _k(q), wherein:

$d_k\left(q\right)>0,\∀k\∈\{1,n_c\}---\left(8\right)$

Wherein d _kthe possible n of the point on the health being included in virtual subject _cto the minor increment between point and another point on the health of virtual subject or the point in another outer body of existing in the design tested.Therefore, when minimizing f (q) all the time, maintain d _k(q) on the health of virtual subject institute a little for be greater than zero.

Generally, according to following formula determination initial attitude:

$\underset{q}{\min }{f}_1\left(q\right)+f_2\left(q\right)$

s.t.q _min≤q≤q _max.

τ _lb≤τ _s≤τ _ub.

$0<d_k\left(q\right),\&ForAll;k\&Element;\{1,n_c\}$

In one embodiment, nonlinear constrained optimization can be used to solve to determine initial attitude.The example that nonlinear constrained optimization solves comprises MATLAB ^tMoPTIMIZATIONTOOLBOX and non-linear internal point trust region optimize (KNITRO).

Posture is determined

Kinematics model

As introduced above, posture certainty annuity 104 is configured to determine the posture of main body when also observing the constrain set applied the motion of virtual subject in the whole process realizing one or more task.In one embodiment, closed form multiple-objection optimization (MOO) technology is used to determine initial attitude.This combine with technique differential kinematics model is to determine posture.This technology is derived and real time execution resolvedly.Use this technology, system 104 is at t ₀initial attitude after multiple time t export posture as orientation vector set q, wherein the number of time t depends on the attitude framework number and task to be done of wishing in posture.

As described above, each vector q comprises the numerical value for each degree of freedom in the radial type model of virtual subject.Can represent that in the attitude of time t be vector q=[q ₁..., q _n] ^t.Here, n represents degree of freedom sum.Indivedual, each vector for respective time t represents the attitude of virtual subject.Generally speaking, vector q represents the posture of virtual subject together with parameter.Such as, if task pulls rim brake, then posture by the joint that represents virtual subject from starting drag rim brake to the orientation (and therefore representing the location in these joints) in the whole motion process completed.

About task, more than one task to be done can be had.Thus, i (i=1 here ... k) be the index associated with each task.Consider the scene for performing k operation task, the location of these operation tasks and/or the time history of orientation are designated.For each task, vector p represents the location of each task and/or the time history of orientation.And virtual subject only uses the angle for each degree of freedom to represent, in contrast, the vector p for task can comprise location (i.e. Descartes) component and rotation (i.e. angle) component.

Posture determines it is kinematics tracking control algorithm, because posture certainty annuity 104 attempts allowing posture tracing task.For task, the space velocity vector associated with the task of specifying is given by following formula:

${\stackrel{\&CenterDot;}{v}}_i={\left[\begin{array}{cc}{w}_i& {\stackrel{\&CenterDot;}{p}}_i\end{array}\right]}^T,---\left(9\right)$

Wherein w _ibe task framework angular rate and descartes's speed of task i.Here, the reference frame of task p is the global coordinate system with reference to initially aiming at the pelvis of virtual subject.Task framework is the reference frame of the body segment associated with task p.The motion of task p can describe relative to the overall reference frame in health outside.Alternatively, the motion of task p can describe relative to global coordinate system or relative to the motion of the pelvis of virtual subject.And not all task will have locating and orienting component.Some tasks only will have positioning component, and some tasks only will have directional component.

In order to determine that radial type model is at t ₀initial attitude after the attitude in moment any time, use differential kinematics model.The Motion mapping needed to realize each task in the task of specifying (such as, by path or use starting point and terminal) is moved to the correspondence of virtual subject by differential motion model.This is realized by degree of freedom (such as joint) value in time changing virtual subject.This speed (or joint space speed) being created in joint to task speed (or task space speed) between mapping.In one embodiment, the differential kinematics model can expressed for performing this mapping is:

$v=J\stackrel{\&CenterDot;}{q}---\left(10\right)$

Wherein J corresponds to augmentation Jacobian matrix,

$J={\left[\begin{array}{ccccc}{J}_1^T& ...& J_i^T& ...& J_k^T\end{array}\right]}^T---\left(11\right)$

Jacobi is the partial derivative of each task p (for k task) relative to q.In other words, it is the motion of the task for the motion in the joint of human subject.Jacobian matrix can be broken down into it respectively by J _oand J _pthe rotation represented and translational component.

$J=\left[\begin{array}{c}{J}_o\\ J_p\end{array}\right].---\left(12\right)$

The inverse of differential motion model is utilized to determine posture

Be used in the differential kinematics model described in equation 10, by determining determine at any given moment time t _lattitude q.When task starts, initial attitude can be used.In one embodiment, by being minimized in time t _lthe task orientation of prediction and/or directed p _lwith vector q _lbetween Descartes's error determine at any given moment time t _lattitude q.Single order closed loop inverse kinematics (CLKI) formula that peer-to-peer 10 is inverted can be used for minimizing Descartes's error and determining attitude q.Add feedback correction term to improve tracking performance.CLIK formula is:

$\stackrel{\&CenterDot;}{q}=J^{+}({\stackrel{\&CenterDot;}{v}}_d+K_{p}e)---\left(13\right)$

Wherein v _dbe the space velocity vector of wishing, and wherein e is the tracking error between the task vector and the task vector of prediction of hope.The attitude of prediction is obtained by the numerical integration of peer-to-peer 13.Once obtain q, the task vector of prediction can by solving Forward kinematics equation to calculate, and these Forward kinematics equatioies are functions of q.Comprise locating and orienting (P _d, Θ _d) the task vector p of hope _lknown from task itself (such as from the motion path provided or terminal).K _pthe diagonal state feedback gain matrix that the rate of convergence of error is controlled, and J ⁺right pseudoinverse by the J of positive definite matrix W weighting:

J ⁺=W ^-1j ^t(JW ^-1j ^t) ^-1(14) in practice, consider the appearance of singular point in matrix J, equation 14 can be replaced by the healthy and strong damped least squares pseudoinverse of singular point.

Tracking error e for individual tasks i can comprise positioning error and orientation error component.These can be expressed as together:

$e_i={\left[\begin{array}{cc}{e}_{o_i}& e_{p_i}\end{array}\right]}^T---\left(15\right)$ Defining positioning error is simply wherein and p _icorrespond respectively to the task orientation of wishing and predicting.Can express orientation error according to angle and axis error is:

$e_o=\frac{1}{2}(n\&times;n_r+s\&times;s_r+a\&times;a_r)---\left(16\right)$

Wherein and R _ithe unit vector ternary of hope and prediction that=[n s a] corresponds respectively to task orientation represents.

Weight matrix W implements the about intrafascicular at least some constraint limited the motion of virtual subject in design.In one embodiment, weight matrix W implements the compound weight matrix that joint constraints constraint, self-gating constraint, the uncomfortable constraint in joint and energy charge retrain (also referred to as dynamic conformance constraint).In one embodiment, compound weight matrix W is the diagonal matrix that its element is derived from constrain set:

W＝aW _h+(1-a)W _f+W _d(17)

Wherein W _hthe weight matrix that its element is derived from the uncomfortable constraint of joint constraints constraint and joint, W _fthe weight matrix that its element is derived from crash restraint, and W _dit is the weight matrix that its element is derived from energy charge constraint.Parameter a is the scalar index that can be used for modulating the contribution of the first two weight matrix.Below further describe each constraint that these are about intrafascicular.

Determine middle in conjunction with contiguity constraint to posture

When assessing design, constrained type is the contiguity constraint between virtual subject and the parts of design.Contiguity constraint instruction virtual subject is anticipated in the whole process of the task of realization and maintains the surface contacted with it.The example of contiguity constraint be virtual subject leg and lean on seat the back of the body, lean on constant contact between the head of resting the head on and the pin leaning on one or more pedal in pedal.

In one implementation, in above equation 13 describe inverse kinematics model in conjunction with contiguity constraint.In this implementation, task to be achieved and contiguity constraint to be observed are considered as segregant task, and there is their priority for realizing each subtask.Contiguity constraint is higher priority subtask.Reality to be achieved (or operation) task is more low priority subtask.In one embodiment, operation in the operation task to be achieved kernel (null-space) in contiguity constraint subtask.When realizing operation task, contiguity constraint cannot be carried out simultaneously mean that this design is infeasible for the virtual subject discussed.The number of the contiguity constraint that can carry out depends on the degree of redundancy of system.Degree of redundancy can based on than in order to realize task and also observe constrain set and the number of degrees of freedom of the less virtual subject of the number of degrees of freedom that needs is determined.To contiguity constraint, priority can be set to give the priority higher compared with another constraint to a constraint in advance or during emulating.

Use the differential kinematics model from above equation 10, contiguity constraint can be expressed as:

$v_c=J_c\stackrel{\&CenterDot;}{q}---\left(19\right)$

Wherein v _cthe rate vector and J that retrain _cit is association Jacobian matrix.In many cases, contiguity constraint is included in the contact point between virtual subject and the parts of design, and wherein contact point is fixing relative to overall framework.In these cases, therefore v _c=0.

Can by represented by formula in conjunction with the kinematics model (equation X) for realizing task and the inverse kinematics model for the kinematics model (equation 10) of observing contiguity constraint:

$\stackrel{\&CenterDot;}{q}=J_c^{+}{v}_c+{\hat{J}}_t^{+}(v_t^{*}-J_t{J}_c^{+}{v}_c)---\left(20\right)$

Wherein:

$\hat{J}=J(I-J_c^{+}{J}_c)---\left(21\right)$

And wherein I is unit matrix and v ^*=(v _d+ K _pe), and wherein as described above, J ⁺=W ^-1j ^t(JW ^-1j ^t) ^-1(equation 14 repeats to know), and wherein:

${J_c}^{+}=W^{-1}{J}_c^T{\left(J_c{W}^{-1}{J}_c^T\right)}^{-1}---\left(22\right)$

Section 1 in equation 22 higher priority subtask for carrying out contiguity constraint is described.Section 2 to drop in the kernel of major subtask and to be included to executable operations task.As previously described, the generalized inverse in equation 20 with by W weighting to meet the constraint except contiguity constraint in constrain set.

As introduced above, at any given moment time t _lattitude q by determining $\stackrel{\&CenterDot;}{q}=q_l-q_{l-1}$ Determine.

Uncomfortable and joint constraints retrains

Uncomfortable constraint is configured to reward following virtual subject attitude, and wherein the joint of main body is close to neutral location, and punishes virtual subject attitude more and more along with the approaching joint constraints in joint of main body.Therefore, uncomfortable constraint (is q for joint i based on the joint of main body _i) limit q relative to the upper angle in each joint _{i, max}) and lower angle restriction q _{i, min}) current angular assessment virtual subject uncomfortable level.According to this point, uncomfortable constraint h ₁q () can be expressed as:

$h_1\left(q\right)={\mathrm{\&Sigma;}}_{i=1}^n{a}_i{\left(\frac{q_i-q_{i,N}}{q_{i,\max }-q_{i,\min }}\right)}^2---\left(23\right)$

Wherein a _ithe zoom factor depending on joint, q _irepresent the generalized coordinate of i-th degree of freedom, and q _{i; N}it is the neutrality location in joint.

By the h represented ₁gradient represent joint constraints gradient function, this function is n × 1 vector, and the entry of this vector is at h ₁q the side advanced the speed the soonest of () is directed upwards towards.

${\&dtri;h}_1=\frac{{\&PartialD;h}_1}{{\&PartialD;}_q}=[\frac{{\&PartialD;h}_1}{{\&PartialD;q}_1},...|,\frac{{\&PartialD;h}_1}{{\&PartialD;q}_n}].---\left(24\right)$

The element associated with joint i is given by following formula:

${\&dtri;h}_{1i}=\frac{{\&PartialD;h}_1}{{\&PartialD;q}_i}={2\mathrm{\&alpha;}}_i\left(\frac{q_i-q_{i,N}}{q_{i,\max }-q_{i,\min }}\right)---\left(13\right)$

If joint is at its neutral pose, then function equal zero, and this function increases towards Restricted Linear.

Although uncomfortable constraint penalty from the neutrality in joint location away from joint motions, it does not implement joint constraints constraint.Joint constraints constraint can be expressed as:

$h_2\left(q\right)={\mathrm{\&Sigma;}}_{i=1}^n\frac{1}{4}\frac{{(q_{i,\max }-q_{i,\min })}^2}{(q_{i,\max }-q_i)(q_i-q_{i,\min })}---\left(2\underset{\&OverBar;}{5}\right)$

Joint constraints function h ₂during restriction in joint close to them, there is much higher value and often in joint constraints for infinitely great.Joint constraints function can be expressed as:

${\&dtri;h}_{2_i}=\frac{{(q_{i,\max }-q_{i,\min })}^2({2q}_i-q_{i,\max }-q_{i,\min })}{4{(q_{i,\max }-q_i)}^2{(q_i-q_{i,\min })}^2}---\left(26\right)$

If joint is at the centre of its scope, then function equal zero, and this function becomes infinity in arbitrary restriction.

The superposition of joint constraints and uncomfortable constraint carried out combining to their individual effects thus prevent joint violate their restriction and punishment from their neutrality location away from joint motions.The combination of two constraints can be expressed as:

h(q)＝h ₁(q)+h ₂(q) (26)

The gradient of composite function h (q) is used for being configured to the weight matrix W used when determining posture of one or more joint constraints and uncomfortable constraint _h.W _hthat there is diagonal element diagonal matrix n × n.Diagonal matrix is calculated by following formula:

$W_{h_i}=\left\{\begin{array}{ccc}1+\left|{\&dtri;h}_i\right|& \mathrm{if}& \mathrm{\&Delta;}\left|{\&dtri;h}_i\right|\&GreaterEqual;0,\\ 1& \mathrm{if}& \mathrm{\&Delta;}\left|{\&dtri;h}_i\right|<0.\end{array}\right.---\left(27\right)$

H _ithis is combination joint for joint i and uncomfortable constraint, gradient, and this represents the change of the magnitude of gradient.On the occasion of instruction joint towards its moving-limiting, and negative value instruction joint is removed from its restriction.During in joint towards its moving-limiting, association weighting factor become very large thus the motion in joint (such as in the q in future time moment determines) in synthesis posture is determined is slowed down.When closely reaching its restriction in joint, weighting factor approaching infinity, and corresponding joint stops in fact in synthesis posture is determined.If joint is removed from restriction, then the motion in joint is unrestricted.

Constraint is avoided in collision

Collision between the different segment (such as arm and chest) that collision avoids retraining the radial type model prevented at virtual subject or between the sections and the parts (such as arm and door) of design of radial type model.If two sections are connected in joint, then can limit range of articulation by using joint constraints constraint thus prevent the collision between two sections.Collision is avoided retraining and is configured to prevent the collision between the sections not sharing identical joint and between sections and parts in the design.Collision avoids constraint to be included in minimum Euclideam distance d (d >=0) between two collisions things (two sections or sections and parts).In one embodiment, crash restraint f (q, d) has minimum value at d=0 and increases along with d and exponentially fail towards zero, such as:

f＝ρe ^-αdd ^-β(28)

Decay rates is controlled by adjustment parameter alpha and β.By increasing α, can control characteristic decay rates, thus this function quickly approaching zero.Parameter ρ span of control limit of control.Be expressed as the gradient of crash restraint f be n × 1 vector, the entry of this vector is directed upwards towards in the side advanced the speed the soonest of f.

$\&dtri;f=\frac{\&PartialD;f}{\&PartialD;q}=[\frac{\&PartialD;f}{{\&PartialD;q}_1},...,\frac{\&PartialD;f}{{\&PartialD;q}_n}]---\left(29\right)$

The gradient of collision can use following formula to calculate:

$\frac{\&PartialD;f}{\&PartialD;q}=\frac{\&PartialD;f}{\&PartialD;d}\frac{\&PartialD;d}{\&PartialD;q}---\left(30\right)$

F relative to the partial derivative of d is:

$\frac{\&PartialD;f\left(q\right)}{\&PartialD;d}=-{\mathrm{\&rho;e}}^{-\mathrm{\&alpha;d}}{d}^{-\mathrm{\&beta;}}(\mathrm{\&beta;}{d}^{-1}+\mathrm{\&alpha;})---\left(31\right)$

D relative to the partial derivative of q is:

$\frac{\&PartialD;d}{\&PartialD;q}=\frac{1}{d}{[J_a^T(x_a-x_b)+J_b^T(x_b-x_a)]}^T---\left(32\right)$

Wherein x _aand x _brepresent orientational vector, the instruction of these orientational vectors is sought by with the Cartesian position retraining two objects (such as sections a and b or sections a and parts b or on the contrary) avoiding it collide.J _aand J _bassociate Jacobian matrix for a with b.Coordinate x _aand x _bcollision detection software can be used obtain.

Each degree of freedom that the element of the vector in the crash restraint gradient function of equation 29 represents virtual subject affects the degree of collision distance.In one embodiment, crash restraint gradient function is used for being configured to according to one or more crash restraint the weight matrix W that uses when determining posture _f.W _fthat there is diagonal element n × n diagonal matrix.Diagonal element is calculated by following formula:

$W_{f_i}=\left\{\begin{array}{ccc}1+\left|{\&dtri;f}_i\right|& \mathrm{if}& \mathrm{\&Delta;}\left|{\&dtri;f}_i\right|\&GreaterEqual;0,\\ 1& \mathrm{if}& \mathrm{\&Delta;}\left|{\&dtri;f}_i\right|<0.\end{array}\right.---\left(33\right)$

? represent the crash restraint gradient function relative to joint i, item the change of the magnitude of representative collision gradient function. make virtual subject mobile towards collision on the occasion of instruction joint motions, and negative value instruction joint motions make dummy model remove from collision.

When the motion in joint makes virtual subject mobile towards collision, the associated weight value factor become very large thus joint is slowed down.When the approaching contact of two limbs, weight approach infinity, and the joint contributed to towards collision movement stops in fact.If the motion in joint makes limbs remove from collision, then the motion in joint is unrestricted.Different crash restraint may reside in the constrain set for seeking each collision prevented.

Dynamic conformance retrains

The understanding moving control from the mankind discloses the mankind can optimize them energy ezpenditure when performing Redundant task.That is, if cartesian space input describe there is the infinite Joint motion configuration space solved, then minimize kinetic energy solve can be predict the nature mankind's attitude time suitable selection.Here, dynamic conformance constraint is configured such that for solving as follows of DOF vector q preferential, and this solves and minimizes (or at least reducing) and be used for required kinetic energy of finishing the work.Also can explain that dynamic conformance is constrained to this and solves and ensure dynamically to balance virtual subject when realizing task.

Can remember that the dynamic of human body in joint space is according to (n × n) joint space inertial matrix (JSIM) H (q):

$\mathrm{\&tau;}=H\left(q\right)\stackrel{\&CenterDot;\&CenterDot;}{q}+C(q,\stackrel{\&CenterDot;}{q})q+\stackrel{\&CenterDot;}{{\mathrm{\&tau;}}_g\left(q\right)}+J^T{f}_e,---\left(34\right)$

Wherein (n × 1) generalized vector of location, joint, speed, acceleration and power is represented respectively with τ.C is (n × n) matrix, thus coriolis and centrifugal item (n × 1) vector.τ _git is gravity item (n × 1) vector.J is Jacobian matrix, and f _ethe space external force acting on system.

The inverse differential kinematics model of solve equation 13 and 20 is similar to the minimum instantaneous weighting kinetic energy determining virtual subject, because a square sum for joint speed is minimized.Determine this of minimum instantaneous kinetic energy to improve, constrain set comprises dynamic conformance weight matrix W _d, this dynamic conformance weight matrix comprises the diagonal coefficient corresponding with the diagonal element of joint space inertial matrix H (q).W _dcan be expressed as:

W _d＝diag(H(q))-I. (35)

Wherein I is unit matrix.In practice, dynamic conformance weight matrix W _dimpact be punishment to following joint motions, this joint motions dislocation has sections, the such as trunk of large quality and/or inertia.

The power being used for generating motion can be provided for the useful information assessing design.In one embodiment, can solve H (q) with determine within the duration of motion (such as the duration of task) activate average absolute power required for all joints to determine the cost of energy of this motion.

$P_{\mathrm{tot}}=\frac{1}{N_S}{\mathrm{\&Sigma;}}_{j=1}^{N_S}\left|{\mathrm{\&tau;}}_j^T\right|\left|{\stackrel{\&CenterDot;}{q}}_j\right|---\left(36\right)$

Exemplary method

Fig. 2 be according to an embodiment for determining to execute the task in vehicle occupant package design time virtual subject attitude in time process flow diagram.Computer system receives 210 at least one task, the virtual subject model comprising parameter, vehicle occupant package design and constrain set.Computer system 100 determines 220 initial attitudes before virtual subject is executed the task.Use initial attitude as starting point, computer system 100 determines the posture in the whole process that 230 virtual subjects are executed the task at virtual subject.Posture determines to comprise the attitude determined for each chronomere in multiple chronomere during virtual subject realizes task.Therefore, this determines that 230 can relate to pragmatic current the attitude being recycled and reused for each chronomere in office and determine.Whether computer system 100 analyzes 240 postures to determine to design is feasible for virtual subject.

Fig. 4 is the process flow diagram for determining indivedual attitude during realizing task according to an embodiment.Fig. 4 illustrates an iteration of the determination 230 from Fig. 2.As described above, attitude and heading reference system 104 accesses 402 from the attitude q of previous time unit t-1 _t-1.Attitude and heading reference system 104 also accesses 404 task p to be achieved.Attitude and heading reference system 104 also accesses 406 constrain sets of will observe during realizing task p.

Attitude and heading reference system 104 uses these inputs to determine that 408 for the attitude q of future time unit t _t410.For the attitude q of time t _tbe collected for exporting to analytic system 106 together with 410 attitudes can determined with other.Additionally, for the attitude q of time t _t410 are fed in posture certainty annuity 104 for determining the attitude q in future time moment _t+1(not shown).Repeat this process until finish the work or until no longer constrain set can be observed.If violate constraint, then replace and return complete posture, posture certainty annuity 104 mistake can exit and indicate task can not be realized by virtual subject.

Additional consideration

Mean that special characteristic, structure or the characteristic in conjunction with the embodiments described is included at least one embodiment to quoting of " embodiment " or " embodiment " in the description, phrase " in one embodiment " or " embodiment " appearance everywhere in the description may not all refer to identical embodiment.

The some parts of embodiment is presented in the algorithm and symbol expression of the operation to the data bit in computer memory.These arthmetic statements and expression are the means that data processing field technician is used for passing on most effectively to this field others skilled in the art the essence of their work.Here and general imagination algorithm be facilitate the result of hope from consistent step (instruction) sequence.Step is the step needing physical manipulation physical quantity.But although this tittle may not adopt the form of electricity, magnetic or the light signal that can store, transmit, combine, compare and otherwise handle usually.Mainly for the reason of common usage, these signals are called that position, value, unit, symbol, character, item, number etc. are easily sometimes.In addition, will need physical manipulation or the conversion of physical quantity or some arrangements of steps of the expression of physical quantity is called that module or code devices are also easily sometimes, and without loss of generality.

But, all these will associate with suitable physical amount with term similar and be only be applied to this tittle facilitate label.Unless as from following discussion clearly another have specifically express, understand action and process that the discussion running through this description utilization such as term of " process " or " calculating " or " computing " or " determination " or " display " or " determination " etc. refers to computer system or similar electronic computing equipment (such as concrete computing machine), this computer system or electronic computing device are handled and conversion stores at computer system memory or register and other such information, transmit or be expressed as in display device the data that physics (electronics) is measured.

Some aspect of embodiment comprises process steps described herein and instruction with the form of algorithm.It should be noted that process steps and the instruction that can embody embodiment in software, firmware or hardware, and these process steps and instruction can be downloaded when embodying in software so that the different platform residing at several operation systems and use to operate these process steps and instruction from these different platforms.Embodiment also can in the computer program that can perform on a computing system.

Embodiment also relates to a kind of device for performing operation herein.This device can be configured to such object particularly, such as concrete computing machine, or it can comprise the multi-purpose computer that the computer program that stores in a computer activates selectively or reconfigure.Such computer program can be stored in computer-readable recording medium, be such as but not limited to comprise floppy disk, CD, DVD, CD-ROM, photomagneto disk any type dish, ROM (read-only memory) (ROM), random access memory (RAM), EPROM, EEPROM, magnetic card or light-card, special IC (ASIC) or be suitable for store electrons instruction and be coupled to separately in the medium of any type of computer system bus.Storer can comprise any equipment in above equipment and/or following miscellaneous equipment, these miscellaneous equipments can store information/data/program and can be transient state or non-transient medium, and wherein non-transient or non-momentary medium can comprise storage/memory information stored more than minimum duration.In addition, the computing machine of indication can comprise single processor or can be use multiprocessor to be designed for the framework increasing computing power in the description.

The algorithm presented herein is not relevant with any certain computer or other device inherently with display.Various general-purpose system also can use together with the program of the instruction according to this paper, or the more specialized device of structure can confirm it is easily with manner of execution step.Structure for these systems multiple becomes clear by the description from this this paper.In addition, with reference to any certain programmed language, embodiment is not described.The instruction and providing understanding multiple programming language and can be used for implementing embodiment as described herein is quoted to realize and optimal mode any of concrete syntax herein.

In addition, the speech used in the description is main for readable and instruct object and selected and can not yet be selected as defining or limit subject matter content.Thus, the disclosure of embodiment is intended to illustrate but not limits the scope of the embodiment set forth in the following claims.

Although illustrated and described specific embodiment and application herein, but be not limited to precise arrangements disclosed herein and parts by understanding embodiment, and various amendment, change and change can be carried out in the layout of the method and apparatus of embodiment, operation and details and do not depart from embodiment as the Spirit Essence that defines in the following claims and scope.

Claims (24)

1., by a computer-implemented method, comprising:

Access radial type system is at the initial attitude of initial time, and described initial attitude comprises the angle of the multiple degree of freedom for articulated system;

The task that access vehicle occupant package design and articulated system will realize in described design;

Access constraints set, how described constrain set restriction handles the described degree of freedom of articulated system to realize described task, and wherein said constraint comprises the contiguity constraint of the motion away from the parts be present in described design preventing articulated system from carrying out; And

Determine the multiple attitudes for radial type system during realizing described task after described initial time,

Wherein determine that described attitude is included in also in accordance with the described degree of freedom manipulation in time realizing described task determining articulated system while described constrain set.

2. method according to claim 1,

Wherein determining that described attitude comprises makes described contact task comparatively realize described priority of task.

3. method according to claim 1,

Wherein said constrain set comprises uncomfortable target, and

Wherein determine that described attitude comprises to be weighted based on the described manipulation of described uncomfortable target to described degree of freedom.

4. method according to claim 1,

Wherein said constrain set comprises joint constraints target, and

Wherein determine that described attitude comprises to be weighted based on the described manipulation of described joint constraints target to described degree of freedom.

5. method according to claim 1,

Wherein said constrain set comprises collision and avoids target, and

Wherein determining that described attitude comprises avoids the described manipulation of target to described degree of freedom to be weighted based on described collision.

6. method according to claim 1,

Wherein said constrain set comprises self-gating and avoids target, and

The described attitude wherein determining one of described follow-up time comprises avoids target to determine to be weighted to the described of the angle of the described renewal of the described degree in described degree of freedom based on described self-gating.

7. method according to claim 1, wherein determine described attitude comprise to from previous time described attitude and perform closed loop the computation of inverse-kinematics based on described constrain set.

8. method according to claim 1, also comprises:

Analyze described attitude to determine whether described operation task can be realized by described virtual subject.

9. method according to claim 1, also comprises:

Analyze described attitude to determine the feasibility of described task, wherein said feasibility is based at least one item come in the group of free the following formation:

The torque that articulated system completes described task and needs,

The power that articulated system completes described task and needs,

The uncomfortable degree that articulated system causes when completing described task,

Complete the amount of energy that described required by task is wanted; And

The quantity of the last physiological effort required for described attitude is kept when completing described task.

10. method according to claim 1, wherein said radial type model belongs to virtual subject human subject being carried out to modeling.

11. 1 kinds of non-transient computer-readable recording mediums comprising executable computer program code, described code comprises the instruction being configured to perform following operation:

Access radial type system is at the initial attitude of initial time, and described initial attitude comprises the angle of the multiple degree of freedom for articulated system;

The task that access vehicle occupant package design and articulated system will realize in described design;

Access constraints set, how described constrain set restriction handles the described degree of freedom of articulated system to realize described task, and wherein said constraint comprises the contiguity constraint of the motion away from the parts be present in described design preventing articulated system from carrying out; And

Determine the multiple attitudes for radial type system during realizing described task after described initial time,

Wherein determine that described attitude is included in also in accordance with the described degree of freedom manipulation in time realizing described task determining articulated system while described constrain set.

12. non-transient computer-readable recording mediums according to claim 11,

Wherein determining that described attitude comprises makes described contact task comparatively realize described priority of task.

13. non-transient computer-readable recording mediums according to claim 11,

Wherein said constrain set comprises uncomfortable target, and

Wherein determine that described attitude comprises to be weighted based on the described manipulation of described uncomfortable target to described degree of freedom.

14. non-transient computer-readable recording mediums according to claim 11,

Wherein said constrain set comprises joint constraints target, and

Wherein determine that described attitude comprises to be weighted based on the described manipulation of described joint constraints target to described degree of freedom.

15. non-transient computer-readable recording mediums according to claim 11,

Wherein said constrain set comprises collision and avoids target, and

Wherein determining that described attitude comprises avoids the described manipulation of target to described degree of freedom to be weighted based on described collision.

16. non-transient computer-readable recording mediums according to claim 11,

Wherein said constrain set comprises self-gating and avoids target, and

The described attitude wherein determining one of described follow-up time comprises avoids target to determine to be weighted to the described of the angle of the described renewal of the described degree in described degree of freedom based on described self-gating.

17. non-transient computer-readable recording medium according to claim 11, wherein determine described attitude comprise to from previous time described attitude and perform closed loop the computation of inverse-kinematics based on described constrain set.

18. 1 kinds for controlling the system of radial type system, described system comprises:

For performing the processor of executable computer program code;

Computer-readable recording medium, comprises the described executable computer program code being configured to perform following operation:

Access radial type system is at the initial attitude of initial time, and described initial attitude comprises the angle of the multiple degree of freedom for articulated system;

The task that access vehicle occupant package design and articulated system will realize in described design;

Access constraints set, how described constrain set restriction handles the described degree of freedom of articulated system to realize described task, and wherein said constraint comprises the contiguity constraint of the motion away from the parts be present in described design preventing articulated system from carrying out; And

Determine the multiple attitudes for articulated system after described initial time during realizing described task,

Wherein determine that described attitude is included in also in accordance with determining while described constrain set that the described degree of freedom the realizing described task manipulation in time of articulated system is indulged.

19. systems according to claim 18,

Wherein determining that described attitude comprises makes described contact task comparatively realize described priority of task.

20. systems according to claim 18,

Wherein said constrain set comprises uncomfortable target, and

Wherein determine that described attitude comprises to be weighted based on the described manipulation of described uncomfortable target to described degree of freedom.

21. systems according to claim 18,

Wherein said constrain set comprises joint constraints target, and

Wherein determine that described attitude comprises to be weighted based on the described manipulation of described joint constraints target to described degree of freedom.

22. systems according to claim 18,

Wherein said constrain set comprises collision and avoids target, and

Wherein determining that described attitude comprises avoids the described manipulation of target to described degree of freedom to be weighted based on described collision.

23. systems according to claim 18,

Wherein said constrain set comprises self-gating and avoids target, and

The described attitude wherein determining one of described follow-up time comprises avoids target to determine to be weighted to the described of the angle of the described renewal of the described degree in described degree of freedom based on described self-gating.

24. systems according to claim 18, wherein determine described attitude comprise to from previous time described attitude and perform closed loop the computation of inverse-kinematics based on described constrain set.